[time-nuts] Wavecrest noise measurement hardware

[SAIDJACK at aol.com]

In a message dated 4/25/2007 11:45:41 Pacific Daylight Time, jmiles at pop.net  
writes:

>As  an aside, how do the Wavecrest machines work?  Do they just run  the
>signal into a low-jitter ADC with a high-quality clock and derive  all the
>timing information digitally, or is the box full of low-noise  tunable
>synthesizers, mixers, filters, and the usual  stuff?



Hi John,
 
Wavecrest's DTS-2070 and DTS-2075 manuals have a pretty decent discription  
including block diagrams of just how they do it, and why their approach is so  
"revolutionary" and so much better than anyone else's.

First off, they do true time-intervall (A to B) measurements with an  
interpolator with 800fs or less resolution.
 
So no phase noise measurement using the NIST mixer setup, or the TSC  
ADC-based cross-correlation approach (which are bandwidth limited by  design)
 
There are no mixers or filters, just fast ECL comparators with some delay  
lines (actual coax cables wound into loops etc), and the usual coarse capture  
(100MHz) and fine capture (ADC working on charge pump).
 
There are no bandwidth limits except the speed of the ECL gates/comparators  
and that's several GHz.
 
With this time intervall data, they do FFT's for phase noise, jitter  
analysis, histograms, and all sorts of other fancy stuff, including analog  
"oscilloscope" mode.
 
bye,
Said



************************************** See what's free at http://www.aol.com.